Smartsupp and the EU AI Act: does your chat widget tell people it's AI?
Smartsupp's Mira is an AI shopping assistant built for ecommerce, and its setup is explicit about humanizing the bot: you type a name, choose a gender, and pick an avatar from a gallery, and that identity displays in the widget header and next to every message Mira sends. That's an unusually direct version of the Article 50(1) tension, the product's own onboarding flow asks you to give an AI a human name and a gendered face before it ever talks to a visitor.
The rule itself is short. Article 50(1) of the EU AI Act requires AI systems that interact directly with people to be designed so users know they're dealing with AI, no later than the first interaction, in a clear and distinguishable way. The exception for cases where it's "obvious" is narrow: a natural-language customer-service bot doesn't qualify just because it has a robot icon. A line in your terms of service doesn't satisfy it either. And it's easy to enforce, because a regulator can open your site and start a chat.
Where the disclosure lives in Smartsupp
- The chat widget header, which displays Mira's configured name and avatar throughout the conversation
- The welcome message, which can be set to greet visitors immediately or only after they message first
- Mira's generated replies, product recommendations and answers drawn from your site and product feed
- The handover point when Mira transfers to a human operator, typically on request or when she lacks enough information
Your Smartsupp disclosure checklist
- Open your storefront in a private window and start a chat: does anything say "AI" before or in Mira's first reply, given her name and avatar are front and center in the header?
- Revisit the name/gender/avatar you chose during setup, a human first name with a photorealistic avatar is the exact combination that needs an explicit AI line to balance it
- Add the AI statement to the welcome message so it's visible before a visitor even sends their first message
- Check both welcome-message-first and message-triggered activation modes, since the disclosure needs to survive either configuration
- Confirm the handover-to-human moment is announced in the conversation, not just reflected in faster or different-sounding replies
- Screenshot the widget header and first-interaction state, and export Mira's name/avatar/welcome-message configuration with a date
Watch out for
Smartsupp built Mira's setup around making the assistant feel like a real team member, a named illustrated character greeting shoppers in the widget header. That's effective ecommerce UX and a near-textbook case of what Article 50(1)'s "obvious from context" exception was not written to cover: a human-presenting figure that never says, anywhere, that it's AI.
Common questions
Mira comes with a friendly designed avatar by default, doesn't that style already read as "obviously a bot"?
Not reliably. Illustrated or cartoon-style avatars still read as a character with a name and gender, not necessarily as "this is AI", the safer assumption is that avatar style alone doesn't do the disclosure work, so keep an explicit line in the greeting.
Mira hands off to us when she's unsure, do we need to relabel anything at that point?
Yes, briefly, let the visitor know a human has joined. It doesn't need to be dramatic, just clear enough that they know who they're talking to now versus a moment ago.